One of the best ways to find out what products are likely to be appearing soon is to ask the component vendors. They can often tell you what requirements are being seen in the market, and have a good idea of what development work is ongoing. Although they can't divulge details about their individual customers, it helps to sense the priorities and focus of the industry.
With this in mind, I spoke with Dr Ebrahim Bushehri, CEO of Lime Microsystems, which has developed an advanced RF transceiver chip specifically designed for small cells. Their first product was commercially available in 2011, targeted at 3G residential femtocells while fulfilling the larger BTS requirements up to picocells. Unlike similar products of that period, it was designed to be adaptable to almost any mobile network frequency from 300MHz up to 3.8GHz and capable of providing both 3G and LTE frequencies and bandwidths.
Where are you seeing most business in the small cell market today?
Despite two customers in production with our device and several design wins, we've not yet seen the expected high volume for 3G. On the other hand, we have seen a lot of interest in LTE, especially in Japan and Korea which are very progressive in the deployment of next generation networks. We are well poised to capitalise on this when volumes of LTE shipments increase.
LTE designs today are not yet fully cost optimised for small cells yet, most of the work having been done for the larger macrocells. There are a few companies exclusively focussed on small cell products who may have gained a short term advantage with this approach.
This year, we have seen a lot of interest in combined 3G and LTE small cells. It's clear that a number of major network operators are driving the requirement for this type of combined solution. Several of the leading baseband chipset vendors support this, and we might expect to see mature commercial products as early as Q1 2013 although it will be a while before they appear in volume in real network deployments.
So who will make these small cells?
I think that initially, the global RAN (radio access network) vendors will introduce them to the market with manufacturing done through the ODMs. As the industry matures, the ODMs and niche players will be able to enter – selling directly to the network operators.
How will the industry deal with the complexity of LTE frequencies and modes?
We saw this problem coming as far back as 2006. It isclear that mobile networks capacity is set to grow, and that a variety of additional spectrum are being introduced globally. That's why we developed our transceiver chip, which is well suited to meet the market requirement in terms of standards and required frequency bands.
With so many different bands being allocated worldwide, vendors need to have almost continuous frequency coverage all the way from around 300MHz up to 3.8GHz. It makes a lot of sense to design in a single component that can be adapted or configured as required.
There are two strong commercial drivers for this approach
1) The price of a silicon device is primarily about volume. If a manufacturer can ship large numbers of the same chips, it can offer lower prices. Market share and size are extremely important.
2) A single design saves time, money and complexity. The small cell vendor can design, test and optimise with one component. This is not a trivial task and requires highly skilled engineers, time and test equipment. Where separate designs are required for different frequencies and standards, this activity is duplicated leading to more cost throughout the product lifecycle. It also impacts the time to market. Warehouse inventory costs (stocking many parts for different markets) can become a major issue.
What design choices have you made?
With the low cost of the 3G device being critical for mass market residential femtocells, we chose a SISO (single input/single output) implementation rather than MIMO (multiple input/multiple output). For the early 4G/LTE products, volumes are lower and the most cost effective option is to use two RF chips to provide the MIMO capability. In the future, when numbers pick up, then a single chip 2x2 MIMO design is commercially quite feasible.
The first 3G+LTE small cells are being designed with completely separate circuit boards, but this will evolve to a single board in the near future. Baseband chip vendors are already running 3G and LTE concurrently on their more powerful chipsets. The output is then separated into two different RF chains and handled by two or more different antennas.
What changes do you foresee to the design in the near future?
I think it would be quite some time before we see a combined 3G+LTE transceiver IC. There is a great deal of complexity and functionality involved in a such as platform to provide concurrent 3G and LTE operation. For example, you would need 2 VCOs (Voltage controlled oscillators) in the same chip for producing the desired frequencies for each of the 3G and FDD LTE transceivers. This is very challenging - having 4 VCOs sharing the same Die - and perhaps an initial option would be to combine two separate Dies in a single package.
There has also been considerable effort between the baseband chipset and RF transceiver vendors to standardise the interfaces used between their respective chipsets. This is an ongoing activity and the parallel multiplex data transfer schemes used today may be upgraded in the not too distant future. A lot of effort has been put into the JESD207 standard with a further upgraded version promised. In any case, we expect a wide adoption of a common interface throughout the industry by this time next year.
